Mathematical Model to Predict Individual Survival for Patients With Renal Cell Carcinoma

Abstract

PURPOSE: To develop a multivariate model and mathematical formula capable of calculating personalized survival for renal cell carcinoma (RCC) patients with clinically available variables. PATIENTS AND METHODS: A total of 477 patients out of 661 undergoing nephrectomy at the University of California Los Angeles between 1989 and 1999 were eligible for evaluation and formed the analyzed cohort for this retrospective study. Time to death was the primary end point assessed. Univariate analysis for 14 to 20 variables was conducted, followed by a multivariate Cox analysis. The variables that provided independent information as to the time of death for metastatic and nonmetastatic patients were coded and incorporated into a function based on the Nadas equation principle. RESULTS: For nonmetastatic patients, the significant variables in the multivariate analysis were Fuhrman’s grade and Eastern Cooperative Oncology Group performance status. For the metastatic patients, Fuhrman’s grade, 1997 classification T stage, number of symptoms, nodal involvement, and immunotherapy were independent predictors for survival. These variables, based on the Cox multivariate regression model, were implanted into an exponential Nadas equation. The expected survival predicted by use of the Nadas equations faithfully describes the actual survival based on Kaplan-Meier curves. CONCLUSION: We have developed mathematical equations for estimating survival after radical nephrectomy for RCC. The resulting formulas are capable of better tailoring survival estimates for a specific patient and are based on widely accepted clinical prognostic variables. On validation with external data, this type of representation can be used as a tool for the determination of personalized prognosis and may be useful for patient education and counseling.